[mips patch rfa] slight tweak to vr5400 "mul" opcode

[cgd@broadcom.com]

[ sorry for bogus empty msg! ]

At Sat, 5 Apr 2003 00:56:21 +0000 (UTC), "Eric Christopher" wrote:
> Right. I correlated that with the 5500 documentation that states that
> the mul instruction "has 5432 compatibility as well as two-operand..."
> or something of the sort.

i didn't look for that particular text in the 5500 docs, but i see
that it seems to have:

* mips32/mip64-style "mul" (called "mul64"), which produces a 32-bit
  result in the dest register, and

* the "traditional" 5432-style "mul" (which is technically "multiply
  and move lo"), which produces a 64-bit result in hi[31..0] and
  lo[31..0] and automatically moves the result to the dest register.

Based on that (and the implementation of the simulator's 5400 "mul"
instruction), i'm thinking that my interpretation of what the opcodes
entries should do is correct.

further, the binutils names for the 5500 multiplication instructions
... don't seem to match the documentation or what people programming
for 54xx and expecting compatibility would want.  But, that's _really_
not my problem.

I guess that, in general, this isn't my problem, so "never mind."  8-)



chris